Who the Subjects Were

In 2010, Joseph Henrich, Steven Heine and Ara Norenzayan published a paper with a title that did most of the work: "The weirdest people in the world?" Their observation was arithmetic to begin with. The overwhelming majority of published behavioural science samples came from Western, Educated, Industrialised, Rich, Democratic societies, and a large share came from a narrower group still, undergraduates enrolled in psychology courses at American universities, participating for course credit.

A sampling complaint on its own would be fair and dull. Every field knows its samples are convenient. What made the paper land was the second half of the argument.

Henrich and his colleagues went looking, in the cross-cultural literature, for evidence on how far these samples generalised. They found that on a striking range of measures (visual perception, spatial reasoning, fairness, cooperation, categorisation, moral reasoning, even the self) WEIRD populations were not merely different from the global average. They were frequently at the extreme end of the observed distribution.

The Müller-Lyer illusion is the example that lands hardest. Two lines of equal length, one with arrowheads pointing in and one out; the second looks longer. It is in every introductory textbook as a demonstration of how vision works. Across societies, the size of the illusion varies enormously, and among San foragers in the Kalahari it is close to absent. The best explanation is that the illusion depends on a lifetime in carpentered environments full of right angles, which trains the visual system to read those junctions as depth cues. What the textbook presents as a fact about human perception is substantially a fact about growing up in rectangular rooms.

The ultimatum game makes the same point about behaviour. One player is given a sum and proposes a split; the second either accepts, or rejects and both get nothing. Western undergraduates typically offer something near half, and reject low offers even at a cost to themselves. This was taken as a window onto human fairness intuitions. Run across fifteen small-scale societies and the offers range from roughly a quarter to well over half, with some groups regularly making hyper-fair offers above half that were then rejected, because in a culture of heavy gift-giving obligation, an over-generous offer imposes a debt.

The behaviour, in other words, was tracking local institutions rather than a species-level fairness module. That is the claim, and it is a much stronger one than "we should test more people".

Two misreadings are worth heading off, because both are common.

The first is that WEIRD results are worthless. They are not. They are perfectly good results about the populations sampled, which are large, and which matter, and for many low-level processes the variation across cultures is small. What is unwarranted is the move from the sample to the word "human" in the title and abstract, made without evidence, as a default.

The second is that the paper argues for cultural relativism about everything. It argues almost the opposite: that the distribution of human psychology is an empirical question, that some things vary wildly and others barely at all, and that you cannot know which is which without looking. That is a call for more measurement, not less.

The response has been partial. Samples are somewhat more diverse, journals ask about generalisability, and large multi-site projects now deliberately spread across countries. But the structural obstacles are as they were: research money, journals and universities are concentrated in exactly the places the paper was complaining about, and a psychology department in Boston will always find it cheaper to test its own undergraduates than anyone else. The critique was accepted rather quickly and acted on rather slowly, which is, on the whole, the normal fate of a correct methodological argument.